Random Forest vs Gradient Boosting (XGBoost)
Problem statement
How would you choose between a random forest and gradient-boosted trees for loan default prediction? A deep tree fits the training data almost perfectly but does much worse on validation data. Some labels are wrong, so the team must account for both overfitting and noise. Calculate ensemble variance, compare the training methods, and explain their sensitivity to label noise.
Show hint
Ask whether the trees are trained independently or each depends on the previous ones.
Relevant roles: ML Engineer, Applied Scientist and Data Scientist.